Longitudinal Impedance Characterization of the SPS MBA-QF Unshielded Pumping Ports Simulations, bead-pull and wire measurements Fritz Caspers, Jonas Ghini and Jose E. Varela Outline Unshielded Pumping Ports Introduction Simulations Measurements Conclusions Unshielded Pumping Ports Motivation: There are 42 unshielded pumping ports (according to layouts) in the SPS. In principle, all of them should have two long damping resistors inside. Of course, their impedance has to be included in the longitudinal impedance model of the SPS. Pumping Ports were already a problem in the past: There are old simulations available. There are old measurements available. Simulations were not straight forward due to the lack of horizontal symmetry (which lowers the cut-off frequency of the beam pipes to 1GHz). In addition, first simulation results did not agree with old simulations. Therefore, a measuring set-up has been put together. Also gives us the chance to gather more data on the effect of the damping resistors. Outline Unshielded Pumping Ports Introduction Simulations Measurements Conclusions MBA-QF Unshielded Pumping Ports f [GHz]Z [k]QR/Q [] GTPML boundary condition had to be used due to a bug. No damping resistors included Resonator model parameters for the biggest three modes. Additional resonances needed to fit the complete curve MBA-QF Unshielded Pumping Ports f [GHz]Z [k]QR/Q [] Element Resistor*TypeNum.f [GHz]Z [k] QR/Q [] Unshielded Pumping Ports * Damping Resistors have not been included in Simulations. This column states whether or not the flange SHOULD have a damping resistor inside (and its type). Originally presented at LIU-SPS BD WG the [ ] 2*LongMBA-QF Eigenmode Simulation Results Due to the lack of vertical symmetry, the cut-off frequency of the pipes is around 1GHz. Wakefield simulation results As everyone knows, eigenmode simulations should not be used above cut-off: Q values are overestimated. R/Q values are also incorrect. Short-circuiting the pipes not only avoids losses due to coupling to propagating waveguide modes but also changes the time-averaged mean stored total energy (source of the R/Q error). MBA-QF Unshielded Pumping Ports f [GHz]Z [k]QR/Q [] Element Resistor*TypeNum.f [GHz]Z [k] QR/Q [] Unshielded Pumping Ports * Damping Resistors have not been included in Simulations. This column states whether or not the flange SHOULD have a damping resistor inside (and its type). Originally presented at LIU-SPS BD WG the [ ] 2*LongMBA-QF Eigenmode Simulation Results Due to the lack of vertical symmetry, the cut-off frequency of the pipes is around 1GHz. Wakefield simulation results As everyone knows, eigenmode simulations should not be used above cut-off: Q values are overestimated. R/Q values are also incorrect. Short-circuiting the pipes not only avoids losses due to coupling to propagating waveguide modes but also changes the time-averaged mean stored total energy (source of the R/Q error). Outline Unshielded Pumping Ports Introduction Simulations Measurements Conclusions Bead-pull Measurements The Set-Up The couplers: Traditional approach creates a TEM mode. The resonances of this mode are very annoying. Damping these resonances may even be worse L Schematic of measurements carried out in 1997 Bead-pull Measurements The Set-Up The couplers: The alternative configuration used here is based on the fact that: TE TM fc [GHz] TE TM fc [GHz] The implementation is very easy but: Requires manipulation of the chamber. In this case, the probe was fixed to the chamber by means of copper tape. To change the coupling, the structure needs to be opened. Not difficult to place the probe to have weak coupling. No TEM mode ! Bead-pull Measurements - Intro Bead-pull measurements can accurately measure the R/Q of a resonance. So far, R/Q values ranging from 5 to 600 have been measured with very good agreement to simulated results. The Qs of the aforementioned measurements ranged from 100 to (200MHz TWC SPS cavity). Damping Resistor f res [GHz] QZ [k] R/Q [] MBA QF Non enamelled HFSSNo WakeNo Eig.No Meas.No 2.5% Meas.Short 2.5% Bead-pull Measurements Damping Resistor f res [GHz] QZ [k]R/Q [] Mode I Sim. EigNo Sim. WakeNo Meas.No 10% *Meas.Long *Meas.2 x Long Mode II Sim. EigNo Sim. WakeNo Meas.No 3% *Meas.Long *Meas.2 x Long Mode III Sim. EigNo Sim. WakeNo Meas.No 5% *Meas.Long *Meas.2 x Long-- * Measurements with damping resistors require special care due to the very low Q. Good agreement in the R/Q values between wake field simulations and measurements. Longitudinal impedance successfully characterized. Very good agreement between simulations & measurements. Unshielded PP RF Leakage There is RF leakage coming from the vacuum seal. Torque [Nm] GHz GHz GHz The Qs of the first and third mode have a high dependence on the clamp tightness. The field maximum for both is close to the vacuum seal discontinuity. (The cutting of longitudinal surface currents is what actually creates the leakage). Unshielded PP RF Leakage There is RF leakage coming from the vacuum seal. Torque [Nm] GHz GHz GHz The Q of the second mode does not depend on the clamp tightness. A field zero is close to the vacuum seal discontinuity. Probe Measurements The set-up beam pipes are not specified in the report. Most likely not the same as we have now. Comparing these values with the recently measured ones: Year 2014Year 1997 f [GHz]Q Q Importance of this comparison comes from: Additional Considerations about the Qs Pumping Ports have long bellows. Both the resonant frequency and the Q of the resonances are influenced by the bellow. Year 2014Year 1997 f [GHz]Q Q 0.5%1300 14% 0.6%1575 5% 1%1100 20% Excentric Bellows Mode IMode IIMode III f [GHz]Q Q Q Compressed Reference Extended The compression of the bellows has a huge impact on the Q of the resonances ( up to 20% ). Wire Measurements Traces look fairly similar. The three main resonances of the structure can be identified in both measurements Once the longitudinal impedance is characterized, I wanted to compare the known impedance model with the one that can be measured by the wire method. Sim. vs Bead-pull &Wire Measurements Damping Resistor f res [GHz]QZ [k]R/Q [] Mode I Sim. EigNo Sim. WakeNo Meas.No 10% Meas. WireNo1.502 (+8.6%)1000 (-22%) (+110%) Meas. Wire LOGNo (-50%) No Mode II Sim. EigNo Sim. WakeNo Meas.No 3% Meas. WireNo1.668 (+10%)834 (-49%) (58%) Meas. Wire LOGNo (-92%) Mode III Sim. EigNo Sim. WakeNo Meas.No 5% Meas. WireNo2.032 (+5.6%)1270 (+16%) (-40%) Meas. Wire LOGNo (-90%) factor 10 factor 13 factor 2 Sim. vs Bead-pull &Wire Measurements Damping Resistor f res [GHz]QZ [k]R/Q [] Mode I Sim. EigNo Sim. WakeNo Meas.No 10% Meas. WireNo1.502 (+8.6%)1000 (-22%) (+110%) Meas. Wire LOGNo (-50%) No Mode II Sim. EigNo Sim. WakeNo Meas.No 3% Meas. WireNo1.668 (+10%)834 (-49%) (58%) Meas. Wire LOGNo (-92%) Mode III Sim. EigNo Sim. WakeNo Meas.No 5% Meas. WireNo2.032 (+5.6%)1270 (+16%) (-40%) Meas. Wire LOGNo (-90%) Sim. vs Bead-pull &Wire Measurements The de-tunning of the resonances seems to be between 5 and 10%. Since it is not a constant value, it is difficult sometimes to identify the measured mode. The de-Qing depends a lot on the impedance of the mode. It ranges from 10 to 50%. The higher the impedance the more de-Qed by the wire. The wire diameter also plays an important role. Errors in the R/Q values are significant: In this particular example: Using the Lumped Element Formula: from +? to -50% Using the Improved LOG formula: from -50% to -90% Larger errors in high impedance modes Error in the Z values: Order of magnitude errors have been found for high impedance modes ( > 50k) using the Improved LOG formula. Similar values are obtained for a 50k and for a 130k impedance resonance Suggest an upper boundary for the measurable impedances with this method. For this particular measurement: Outline Unshielded Pumping Ports Introduction Simulations Measurements Conclusions Conclusions (I) The longitudinal impedance model of an MBA-QF unshielded pumping port was presented. Due to the lack of symmetry, wake field simulations were used to compute the impedance. Bead-pull measurements were carried out to confirm the longitudinal impedance of the three biggest resonances of the structure. Very good agreement was found for the R/Q simulated and measured values. Bead-pull measurements continue to prove their effectiveness in measuring the R/Q of a resonance. So far it has successfully measured: SPS 200MHz TWC Closed SPS flange (with and without damping resistor) Enamelled SPS flange (with and without damping resistor) Unshielded pumping port Conclusions (and II) It is not generally possible to determine the longitudinal coupling impedance from the measurement of the S21 if the DUT is considered a black box. The fact that different formulas have to be used for lumped and distributed impedances is a clear indicator for this. E. Jensen, PS/RF/Note Personal Conclusions on Wire Measurements For the considered example, even having accurate knowledge of the impedance of the structure it was difficult to identify the resonances due to the de-Qing and de-tuning effects. In my opinion, wire measurements are not a reliable way of measuring the longitudinal impedance of a structure if not backed by trustworthy simulations. Not even to get an order of magnitude idea of the impedance of the structure. Very similar values (18, 28 and 34k) were obtained for very different resonances (11, 131 and 50k) using the lumped element formula. Very similar values (4.2, 5.2 and 5.6k) were obtained for very different resonances (11, 131 and 50k) using the improved log formula. Sometimes, there is no alternative to a wire measurement. Certain structures can not be accurately simulated due to their complexity and/or the presence of special materials.